About
Gowrishankar — AI Solution Architect
How I think about architecture, and how I got here.
I’m Gowrishankar Sivagnanam, an ai solution architect focused on turning AI capability into systems that hold up under real production load — real users, real latency budgets, real failure modes, and real audit requirements.
Most of my work sits at the intersection of three things: cloud and platform architecture, applied AI/ML (increasingly LLM-based systems), and the organizational work of getting a design actually adopted by the teams who have to live with it. A good architecture that nobody can operate isn’t a good architecture.
I write about the parts of this work that don’t make it into the slide deck — retrieval quality, evaluation harnesses, cost-per-request math, and the trade-offs behind decisions that looked obvious in hindsight but weren’t at the time.
How I approach architecture
Architecture is a series of reversible bets
I optimize for cheap-to-change decisions over theoretically perfect ones. Most systems don't fail from a bad component choice — they fail from an irreversible one made too early.
If it isn't measured, it isn't architecture
Every design decision ships with the metric that will tell us if it was right — latency, cost per request, eval score, incident rate. Opinions don't get to grade themselves.
AI systems still obey systems thinking
LLMs are a new component, not a new discipline. Retrieval, caching, observability, failure isolation — the fundamentals of good architecture apply, and skipping them is how RAG demos fail in production.
Experience
2025 — Present
AI Solution Architect · SRS Global Technologies
Own the architecture for mConsent 2.0 end to end — a strangler-fig migration of a legacy PHP EMR onto Next.js/NestJS microservices, with RabbitMQ for service communication and BullMQ/Redis driving the audit-logging pipeline — designed and built the infrastructure from scratch while leading a small team through it. Also architected API Central, a WebSocket-based remote EMR sync agent, and LeadOS, an AI-enrichment platform for lead generation. Led the team's shift to an AI-native workflow — Cursor plus custom agentic skills spanning requirement capture through commit — that helped ship a two-year roadmap well ahead of schedule; that combination of architecture ownership and AI-adoption leadership is what the title now reflects.
2024 — 2025
Software Engineer · Niotact Digital Solutions
Built Smartyard, an intelligent shipyard-management web and mobile app, and led development of a chemical-plant digitization platform with ETL pipelines and Grafana-based BI dashboards.
2022 — 2024
Software Engineer · Hyniva Consulting
Shipped the Victory Capital web and mobile platform on React/React Native with a serverless AWS Lambda backend, alongside part-time work building a warehouse-management system for Nativitas Consulting.
2021 — 2022
Graduate Engineer · Bass Construction
Started in construction project engineering, then found software while building tools to automate the team's repetitive processes — the pivot that led into full-time development.
Focus areas
A skill tree, not a chip list
Tap a node to unlock its skills.
The stack
A tech shelf, not a logo wall
Every tool below is something I’ve actually shipped with — not a logo cloud copy-pasted from a template.
AWS Certified Cloud Practitioner
AWS
Business Analytics Specialization
IIIT Bangalore